Subaqueous Delta of the Ganges-Brahmaputra River System

ELSEVIER Marine Geology 144 (1997) 81-96

Subaqueous delta of the Ganges-Brahmaputra river system

Steven A. Kuehl a,*, Beth M. Levy a, Willard S. Moore b, Mead A. Allison ’ a Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, VA 23062, USA b Department of Geological Sciences, University of South Carolina, Columbia, SC29208, USA ’ Department of Oceanography, Texas A&M University, 5007 Avenue U. Galveston, TX 77551. USA Received 19 July 1996; accepted 16 June 1997

Abstract

The Ganges-Brahmaputra is among the world’s three largest river systems in terms of sediment load, but, until now, no high-resolution seismic data have been obtained to document the nature of the sediment deposit seaward of the rivers’ mouths. The other two (Amazon, Huanghe) discharge into energetic coastal environments and form subaqueous deltas with characteristic clinoform stratigraphy. High-resolution seismic reflection profiles of the Bengal shelf reveal similar stratigraphy: topset beds dip gently (0.036”) and diverge offshore; more steeply dipping foreset beds (0.190”) converge farther seaward; and relatively thin, gently dipping bottomset beds (0.022”) extend across the outer shelf, overlying an erosional surface presumed to be of Late Pleistocene age. Sediment accumulation rates are highest in the foreset region (2 5 cm/year) and reduced in the bottomset region (co.3 cm/year), corroborating the relative thickening and thinning of strata observed in seismic profiles. Taken together, these data indicate a subaqueous delta is actively prograding across the Bengal shelf. Volume estimates for the Holocene subaqueous delta reveal that about one third of the total load of the Ganges-Brahmaputra has accumulated on the shelf. The remainder is likely partitioned between the river floodplain/delta plain and off-shelf transport via the submarine canyon, Swatch of No Ground. The canyon incises the shelf in the area of highest sedimentation rates (foreset), and growth faults and slumping of modern sediments near the head of the canyon support the idea that significant off-shelf transport of sediments to the Bengal Fan is occurring. 0 1997 Elsevier Science B.V.

Keywords: delta; shelf sedimentation; seismic reflection profiling; sediment budget; Bay of Bengal; Ganges River; Brahmaputra River

1. Introduction delta, commonly are found in quiescent seas or protected locations such as fjords and embay- Modern river deltas have a range of characteris- ments. Rivers entering energetic marine environ- tic morphologies that, to a first approximation, ments display a variety of morphologies. For are controlled by the fluvial, tidal, and wave regime example, the Columbia River (Northwest coast, (Wright and Coleman, 1973). Deltas with extens- U.S.A.) has no subaerial delta, rather, accumula- ive subaerial expression, such as the Mississippi tion occurs as a mid-shelf mud deposit (Wright and Nittrouer, 1995). Rivers with sediment loads * Corresponding author. Fax: + 1 (804) 684 7250: comparable to the Ganges-Brahmaputra, such as e-mail: [email protected] the Amazon and Huanghe, exhibit predominant

0025-3227/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved PII SOO25-3227(97)00075-3 or partial subaqueous growth of their deltas, seaward of the Ganges-Brahmaputra river system. respectively (Nittrouer et al., 1986; Prior et al.. In addition, sediment cores collected from the shelf 1986; Alexander et al., 1991 ). Even though the are used to examine the distribution of recent sediment discharge of the Amazon is sufficient to accumulation rates in relation to seismic observa- produce a sizable subaqueous delta (clinoform) on tions. The objectives are to examine the idea that the shelf, large shear stresses, generated primarily a major component of the Ganges-Brahmaputra by tides, appear to prevent or limit significant delta occurs on the shelf as a subaqueous clinoform subaerial growth in the vicinity of the river’s mouth and to evaluate the role of the Swatch of No (Kuehl et al., 1986; Geyer et al., 1996). For the Ground as a potential conduit for bypassing Huanghe, about lo--15% of the sediment discharge modern river sediments to the Bengal Fan. accumulates south of the Shangdong Peninsula as a subaqueous delta (Alexander et al., 1991), with much of the remaining discharge contributing to 2. Background rapid subaerial growth. This dual-mode progradation of the delta reflects the phasing of river .?.1. River churucteristics urd Holocvnr discharge and energy conditions in the Gulf of .strutigrctph~~qf thu delta Bohai; high discharge occurs during low-energy conditions resulting in deposition near the mouth The Ganges, Brahmaputra and Meghna rivers and significant subaerial growth (Wright and have occupied and abandoned numerous courses Nittrouer, 1995 ). during the Quaternary and have deposited a large. Although the combined sediment discharge of flat, low-lying alluvial/delta plain encompassing the Ganges-Brahmaputra river system is among most of the country of Bangladesh (Coleman, the world’s largest and its delta plain among the 1969). The Ganges drains the south slopes of the world’s most densely populated, little is known Himalayan mountains whereas the Brahmaputra regarding the Holocene evolution of the delta or mostly drains the north slopes, with estimated the processes and patterns of recent deltaic sedi- suspended sediment loads of 520 x lo6 t/year and mentation. Based on examination of nautical 540 x lo6 t/year, respectively (Milliman and charts dating back some 200 years, Coleman Syvitski, 1992). The Meghna River drains north- ( 1969) suggested that no significant seaward pro- eastern Bangladesh but has a negligible impact on gradation of the shoreline had occurred and that the combined load of the system, contributing sediments discharged to the coastal ocean therefore only about 1% of the total sediment discharge escaped to the deep sea through the Swatch of No (Coleman, 1969). Combined monthly sediment Ground, a major submarine canyon feeding the and water discharge reaches a maximum in August immense Bengal Fan. A recent critical examination during the southwest monsoon, with minimal levels of a more extensive set of historical charts. how- (an order of magnitude less than peak rates) in ever, has provided evidence for some recent growth January-March. Suspended sediments of the of the subaerial delta, but with progradation occur- Ganges-Brahmaputra are coarse relative to other ring in a lateral (west to east) fashion (Allison. large river systems; at the confluence of the Ganges 1997). Preliminary study of the continental shelf and Brahmaputra about 40% is sand (mostly seaward of the rivers’ mouths revealed significant fine-very fine) and 60% silt-clay (Barua et al.. sediment accumulation, leading to the suggestion 1998). that the observed clinoform-like morphology Umitsu ( 1985, 1987, 1993) divided the land- reflects the presence of an active subaqueous delta forms of the Bengal Basin into two geomorphologi- (Kuehl et al., 1989). However, until now, no cal units: Pleistocene terrace uplands and recent seismic data have been obtained to document the alluvial lowlands (Fig. 1). The Pleistocene terraces stratigraphic nature of this feature. Here we report are found in the marginal and interior portions of the results of the first high-resolution seismic the Basin. In the north the uplands are known as reflection study conducted on the Bengal shelf the Barind Tract. and as the Madhupur Terrace S. A. Kuehl et al. / Marine Geology 144 (19971 81-96 83

WE 89” 90” 91" 92” 26”N 26”N

25’ ‘25”

24O 24”

23” 23’

22”

21”

8&E QiY

Fig. 1. Physiographic map of the Ganges-Brahmaputra delta.

in the central region. Alluvial lowlands are distrib- last glacial maximum (Umitsu, 1993). The lowest uted widely over the Bengal Basin with characteris- unit dates from the last glacial maximum and tic levees, point bars and channel bars, as well as consists of sandy gravels deposited by the rivers lower landform units such as swamps, marshes down cutting older surfaces. Radiocarbon dating and former river channels. Sediments consist and similarities in sediment facies and grain size mainly of sand, silt, and clay layers with peat of the lower unit near Khulna City with that of layers recognized in several places. The elevation the present Brahmaputra River flood plain suggest of the lowlands typically is < 15 m above sea level that this unit represents the flood plain about and most of the southern region is < 3 m above 12,000 years before present. The middle unit is sea level. composed of fine deltaic sediments deposited Five stratigraphic units characterize the evolu- during the transgression. Peat layers are found in tion of the Ganges-Brahmaputra delta since the the Sylhet Basin in the lower horizon of the unit, indicating the presence of marshes during this western region. Spring winds. combined with the period. The lower portion of the upper unit (mid- Coriolis effect, result in the movement of surface Holocene) is comprised of silt and clay in the water away from the east Indian coast with deeper inland region of Tangail; however, near the more water upwelling, causing the isopycnals to tilt coastal region of Khulna, the sediments exhibit a upward towards the Indian coast. In the autumn. strong marine influence. Umitsu ( 1993) proposed the reverse occurs as water is piled up in the that the coastline at that time retreated slightly western part of the bay and the isopycnals tilt north of the present Khulna City. The upper down toward the east Indian coast. Resulting portions of the upper unit become coarser. with seasonal changes in sea level exceed 1 m for the peat in places, suggesting that the coastline pro- northeast coast at Chittagong and along southeast graded during this time as broad marshy peat Bangladesh, the largest on record (Murty et al.. lands covered the central Ganges-Brahmaputra 1992). Observations and modeling of the fresh delta. During the late Holocene the rate of sea- water emanating from the Ganges-Brahmaputra level rise decreased, allowing fine silts and clays rivers’ mouths indicate that during the period with intermittent peat layers to be deposited, here of maximum sediment and water discharge called the uppermost mit. (June-September), the plume trajectory is along the coast toward the west (Shetye et al., 1996). 2.2. Oceanogruph?! qf’the Ba??qf’Benguqul Tides in the coastal waters off Bangladesh primarily are semi-diurnal. The combined effects of The Ganges-Brahmaputra river system dis- Coriolis acceleration and the funnel shape of the charges into an energetic marine environment char- bay produce an area of increased tidal amplitude acterized by strong tidal currents, moderate wave along the eastern coast, typically about 4 m. activity, seasonal monsoons and frequent cyclones. decreasing to less than 2 m for the western portion The mixing and subsequent spreading of the fresh of Bangladesh. For a small area between Hatia water greatly affects the oceanography of the and Sandwip channels amplitudes of up to 6 m coastal waters. In addition to perennial discharge (Barua et al., 1994) and velocities exceeding from the Ganges and Brahmaputra, seasonal dis- 300 cm/s (Coleman, 1969) are observed. charges are introduced into the Western Bay from Cyclones are common in the Bay of Bengal, and India via the Godavari, Krishna and Cauvery about 16% of cyclonic storms developing in the rivers (Suryanarayana et al., 1992). The advection bay strike the Bangladesh coast (Mooley and and mixing of these fresh waters (about 2300 Mohile. 1983). The frequency distribution of km3/year) with the coastal ocean is controlled by cyclone activity is distinctly bi-modal with peak external and internal forcing from wind and activities in May and October. corresponding to thermohaline conditions (Suryanarayana et al.. the transition periods between the northeast and 1992). southwest monsoons. Seasonal low-pressure areas over the Persian Gulf during the summer and high pressure over 2.3. Sedinzent dispersal on the Bengul shelj’ the Tibetan Plateau during the winter create monsoonal winds, from the southwest in summer and Few published studies have addressed recent from the northeast in winter (Murty et al., 1992). sediment dispersal on the Bengal shelf. Barua et al. Changing monsoonal winds affect surface water ( 1994) show that the magnitude and distribution flow in the open Bay of Bengal (Wyrtki. 1973 ). of suspended sediments during the low-discharge The spring is characterized by clockwise rotation, period are primarily a function of tidal energy in with the fastest flow close to the central Indian the nearshore region (< 15 m water depth). continental shelf, where it can reach 1.552.5 m/s. Sediments which are i 125 urn (fine sand to clay) The autumn is characterized by counterclockwise are continuously transported in suspension, except movement with lower speeds in the eastern and for brief periods during slack water. Based on central regions of the bay as opposed to the mineralogical investigation of surficial sediments, S. A. Kuehl et al. / Marine Geology 144 (1997) 81-96 85

Segall and Kuehl (1992) suggest that only during intervals, which were homogenized and bagged for high-discharge periods (May-October) are signifi- sedimentological and geochemical analyses. A cant amounts of sediment transported seaward of wood fragment recovered from the kasten-core the 20-m isobath. Seabed textural and geochrono- nosepiece at Station 1 (- 1.4 m depth in seabed) logical data reveal a westward fining along the was saved for radiocarbon dating. mid-shelf area with a corresponding increase in Navigation data from the 1991 cruise was con- sediment accumulation rates, suggesting westward verted from local Decca coordinates to latitude transport along the shelf toward the Swatch of No and longitude. Using a 3-point moving average, Ground (Kuehl et al., 1989). horizontal distances (in km) were determined along transects between each time mark, typically every 15 min. Vertical scales were calculated based 3. Methods on the recorder setting, assuming a velocity of 1500 m/s. The seismic profiles were collected in 3.1. Field methods and seismic interpretation analog format using an EPC@ graphic recorder, and key sections were digitally scanned at 400 dpi A total of 450 km of GeoPulse@’ multifrequency resolution. Scanned sections were cleaned, scaled, seismic reflection data were obtained during a 199 1 and mosaicked using Intergraph Microstation@ survey of the Bengal shelf along eight transects and IRASB@ vector/raster processing software. (Fig. 2), with maximum seabed penetration in excess of 100 m and vertical resolution of - 0.5 m. 3.2. Laboratory methods Seabed sampling retrieved nine kasten cores (3-m maximum length) and eleven grab samples. Analyses of short-lived radioisotopes were per- Sediment cores were subsampled by taking centi- formed using gamma spectroscopy. ‘i”Pb and meter-thick sections, typically at 5- or lo-cm i3’Cs activities were determined by direct measure-

Fig. 2. Bathymetric chart of the Bengal shelf showing coring station locations (numbers) and seismic transects (letters). A major submarine canyon, ‘Swatch of No Ground’, incises the shelf along the western side of the study area. ment of their characteristic gamma-ray emissions, whereas 226Ra activities were determined indirectly through measurement of its short-lived daughters. Samples were packed in petri dishes (6.5 cm diame- ter by 2.25 cm), sealed to prevent loss of 222Rn (an intermediate daughter between 226Ra and its measured daughters, 214Pb and 214Bi) and to allow for ingrowth before counting. Self absorption cor- rections for *rOPb were made using the methods of Cutshall et al. (1983). Radiocarbon analysis was performed on the wood sample using the benzene synthesis and liquid scintillation counting method similar to that described by Polach and Stipp ( 1967). Textural analyses were performed by wet sieving samples through a 63 pm sieve to separate the sand and mud (clay and silt) fractions. Sand was sieved at l/2 4 intervals, and detailed analysis of the mud fraction was performed using a Sedigraph@ model 5100 ET X-ray digital settling analyzer. Results of Sedigraph@ and sieve analyses were combined to calculate grain-size and sorting statistics.

4. Results

4.1. Seismic prqfiles

The GeoPulse@ records reveal clinoform stratigraphy for the sediment wedge off the Ganges- Brahmaputra rivers’ mouths (Fig. 3). The clinoform thickness exceeds 60 m on the middle shelf. thinning to < 10 m at the seaward limit of the profiling. Five regions of the shelf have been delineated based on regional variations in acoustic character of the seabed and are described below.

4.1.1. Neurshore (5- I5 m) Seismic profiles from the nearshore region of the Bengal shelf reveal a highly reflective sediment surface with limited acoustic penetration (~25 m) and few distinct subsurface reflectors (Fig. 4). characteristic of high sand content. In the north- eastern section of the bay, along the Chittagong coast, asymmetrical sand waves ranging in height from 3 to 5 m (Fig. 5) are present in the channel (Fig. 2) which presumably is related to the struc- S.A. Kuehl et al. /Marine Geology 144 ( 1997) 81-96 87

1 km I 1 0 WEST EAST

Fig. 4. GeoPulse@ record from the near-shore (transect B) revealing poor acoustic penetration, probably a result of the high abundance of sand in this region. The thin (2-3 m) surface unit may represent a veneer of mud (probably ephemeral) overlying sand typical of the inner topset region.

gE WEST ,lkm , EAST aa > 30 sl: 8; 45

Fig. 5. GeoPulse@ record from the eastern region of the study area (transect G) showing asymmetrical sand waves indicating transport toward the west. Grab samples from this area reveal clean medium sand.

tural troughs of the adjacent Chittagong hill tracts. The inner shelf dips gradually seaward in this The sand waves appear to be more widely spaced region, with an average gradient of 0.036”. No toward the west, in the direction of the lee face. significant change in this gradient is observed from Very shallow (< 15 m) acoustic penetration was east to west. Farther seaward, in water depths of achieved in this area and field descriptions of - 30 m, greater than 30 m of subbottom penetra- surf%zial grab samples indicate a well-sorted tion is achieved and previously-parallel reflectors medium sand. begin to diverge (Fig. 3). The spacing between reflectors increases from ~2-3 to - 5 m with 4. I .2. Inner shelf ( 15-30 m) distance from shore. Closely spaced acoustic reflectors (c 2-3 m spacing) are present in the shallow seabed (upper 15 m) 4.1.3. Middle sheCf (30-60 m) for much of the inner-shelf region. These reflectors Densely spaced reflectors (3-5 m spacing) are are underlain by an irregular erosional surface in observed in the middle-shelf region where typical transects A and C that has an area1 extent of at acoustic penetration between 30 and 45 m is least 720 km2. The erosional surface displays cut achieved. The beds begin to converge seaward in and fill features perhaps associated with subaque- water depths of about 60 m (Fig. 3), where reflec- ous distributary channels (Fig. 6). tor spacing is reduced to < 2-3 m. Individual beds

1 km SOUTH

Fig. 6. GeoPulse@ record from the inner shelf (transect C) revealing an erosional surface beneath thin topset beds. The cut and fill features could reflect migration of subaqueous distributary channels. are difficult to discern in the seaward portion ot similar water depths. On the western side of the the middle shelf and irregular, erosional surfaces canyon, a well-stratified seabed is observed, con- can be seen below the stratified sediments. The taining distinct, parallel, and continuous acoustic gradient of the sediment wedge over the middle reflectors with typical spacing of 3-5 m between shelf averages 0.19”. This gradient appears to be beds. On the eastern side of the canyon, beds are uniform from the eastern transect to the Swatch also densely spaced but are more irregular in of No Ground. nature than their western counterparts. Two distinct acoustically transparent layers, Abundant growth faults and slumps are found about 3-5 m thick, are observed both from shore on the eastern flanks of the canyon. Slumps with perpendicular and shore parallel transects. Located rotational movement to the west are observed at depths of Y 10 and _ 20 m below the sediment along the margin of the canyon (Fig. 9). An surface, the layers are roughly parallel to one east-west trending gully feeds into the canyon with another and to the seabed surface ( Fig. 7). The growth faults along the south side dipping steeply acoustically transparent layers pinch out in water into the canyon (Fig. 10). Near the canyon head, depths of about 80 m on the outer shelf. Based on strata become more irregular and often are chaotic, the seismic data, the layers appear to have an area1 with discontinuous and truncated beds. Profiles of extent of y 1500 km2, extending at least 12 km the canyon floor near its head show irregular strata north-south and 127 km east-west. with few discernible beds and an uneven sediment surface. 4. I. 4. Outer she& ( 60- > 80 m i Profiles from the outer-shelf region reveal closely 4.2. Sediment accumdution rutes spaced (< 2-3 m spacing). parallel acoustic reflectors extending to water depths of at least 80 m, Sediment accumulation rates were estimated pri- the seaward extent of the seismic lines. The seafloor marily using 13’Cs and, in one instance, corrobo- gradient in this area is very gentle, 0.022”. An rated by 210Pb geochronology. In most cases, erosional contact is seen below the shallow strata ‘l”Pb profiles could not be modeled as steady-state relatively close to the seabed surface (within accumulation because activities fluctuate markedly _ 10 m). The erosional contact is uneven and down core, in part a result of grain-size variations displays varying relief, with distinct cut-and-fill ( Kuehl et al., 1989). In these cases, the penetration features (Fig. 8). Strata underlying this reflector depths of 13’Cs were used to provide a first-order are discontinuous with indistinct bedding. estimate (Table 1). As 13’Cs has been present worldwide in measurable quantities since about 4.1.5. Suatch qf’No Ground 1954, the depth to which 13’Cs is observed in the Profiles near the Swatch of No Ground ( Figs. 9 cores divided by the number of years since intro- and 10) reveal increased acoustic penetration rela- duction (in this case 37 years) provides a rough tive to the eastern and nearshore shelf regions of estimate of the sediment accumulation rate. A

1 km WEST EAST

Fig. 7. GeoPulse@ record from the foreset beds (transect F) showing two distinct acoustically transparent beds at about 48 m and 62 m. The _ 5-m thick transparent beds extend at least 100 km along the foreset region and probably represent large-scale mass movement, perhaps triggered by cyclones or earthquake activity common to the area. S.A. Kuehlet al. 1 Marine Geology 144 (1997) 81-96

1 km

Fig. 8. GeoPulse@ record from the seaward portion of transect C, showing thin parallel bottomset beds overlying Pleistocene strata below about 100 m. Cut and fill features are evident in the Pleistocene strata suggesting former channels or gullies. potential problem with this approach is the effect sediment accumulation rates were derived using of seabed mixing (physical or biological) on both 137Cs and ‘l”Pb profiles, with a maximum 137Cs profiles, which would increase penetration rate of 0.3 cm/year. Although some evidence of depths and lead to an overestimate of sediment bioturbation was noted at this station (indicating accumulation rate. Areas most susceptible to this the reported rate may be a maximum), the accumu- effect would be those experiencing low accumula- lation rate was the lowest measured in this study. tion rate (< - 1 cm/year) and/or deep intense Radiocarbon dating of a wood fragment collected mixing. Previous studies have shown that physical from the nosepiece at Station 1 gave an age of sedimentary structures dominate the inner and 10,000 f 240 year. middle shelf area (Segall and Kuehl, 1994), indicating that biological mixing is unlikely to affect 4.3. Grain-size analysis 137Cs penetration depths in these areas. Deep physical mixing, such as that observed for the Textural analyses were performed on samples Amazon delta (Kuehl et al., 1995), would most from six cores (Table 2). Cores from the western likely be a problem in shallow water depths middle shelf (Stations 4, 5 and 6) are very fine- (< -20 m) where the effects of waves and tides grained, with down-core average means of 8.5 4 are most pronounced. Most of the cores examined for Stations 4 and 5, as compared with an average for this study were collected in water depths mean of 7.0 $ for Station 8 on the central middle > 20 m (Fig. 2). Using 137Cspenetration depth as shelf. The cores from the outer shelf (Station 1) a first-order estimate, a sediment accumulation and from the shallow trough in the eastern inner rate of 1.8 cm/year is obtained for a core collected shelf (Station 9) have intermediate average means in a shallow trough in the eastern part of the inner of 7.9 4 and 7.6 $, respectively. shelf (Station 9). This rate could represent a minimum because 137Cs is present throughout the core, thus the actual first appearance of 137Cs in 5. Discussion the seabed may be deeper. Kasten cores could not be obtained from the nearshore region because of 5.1. Subaqueous delta the presence of sands, which hinders gravity coring. For the middle-shelf area, sediment accumulation A distinctive morphology can be seen in the rates were 1.1 cm/year (Station 6) and 0.6 cm/year sediment wedge prograding southward and west- (Station 4). The highest sedimentation rates are ward from the rivers’ mouths on the inner Bengal found in the area surrounding the Swatch of No shelf (Fig. 11). The seismic profiles reveal the Ground. Station 5 revealed a minimum rate of gently sloping topset, more steeply dipping foreset, 5.2 cm/year, based on the presence of 137Cs and gently sloping bottomset stratigraphy charac- throughout the core. For the outer shelf Station 1, teristic of subaqueous deltas (Fig. 3). On the inner 6c- 9c - WEST EAST 1 kn? 120 -

150 -

I80 -

21c -

240-

270-

300-

330-

390-

420-

4!5@---

Fig. 9. GeoPulse@ record from the head of the submarine canyon, Swatch of No Ground (east-west leg of transect E). A series of four large (-0.5 km) slump blocks are seen at the edge of the shelf. Reflectors within the slump blocks dip east. back towards the shelf. Below the shelf edge the profile crosses a deep (-60 m) gully before falling off to the canyon floor. A hummocky sediment surface on the canyon floor suggests mass movement of sediment from the shelf into the canyon. NORTH SOUTH 75 1 Irm

105

120

135

150

165

180

195

210

225

Fig. 10. GeoPulse@ record across a gully on the eastern side of the Swatch of No Ground (northeast-southwest leg of transect E). A series of growth faults is present along the south edge with shear planes dipping down into the gully with decreasing angles towards the south. Deep subbottom acoustic penetration (z 100 m), along with the growth faults and deformed strata at depth, suggest rapid accumulation of fine-grained sediment. The hummocky sediment surface of the gully floor suggest mass movement into the gully/canyon system. Table 1 Table 2 ‘3’Cs-based accumulation rateb Grain-size data

Station Core length 13’Cs penetration Accumulation Sample Mean Mean grain Average mean Sorting Nr. (cm) (cm) rate (cm/year) depth (cm) size (4) for core - I 81 II 0.3 Cow I J I51 21 0.6 A91 1-O 1.0 6.9 7.9 3.5 5 191 I91 > 5.2 A91 l-20 21.0 8.1 2.4 6 so 40 I.1 A91 l-30 31.0 6.5 4.0 9 66 66 > 1.x A91 l-80 81.0 7.6 2.4 A911-100 101.0 8.7 2.6 A91 l-140 141.0 9.5 2.6 shelf, in water depths of about 15-25 m, reflectors Cvrr 4 are nearly parallel and slope gently seaward. This A9 14-8 9.0 6.9 8.5 2. I trend gives way to divergent beds at the seaward A914-15 16.0 9.8 1.8 edge of the topset region, evidenced by increased A914-25 26.0 8.3 _.15 A914-55 56.0 8.2 2.2 spacing between reflectors, indicating increased A914-65 66.0 8.4 3. I sedimentation rates. At depths of -30 m the A914-90 91.0 8.3 2.5 seafloor gradient increases abruptly (from 0.036” A914-140 141.0 9.1 2.3 to 0.19’ ) and reflectors converge seaward across A914-160 161.0 8.8 2.4 the foreset region. In the bottomset region (> 60 m A914-200 210.0 8.2 2.7 water depth) near-surface reflectors once again C’orr 5 take on a semi-parallel appearance with strata that A915-0 1.0 8.7 8.5 3.0 are closely spaced and thin. A hiatal surface is A915-20 21.0 9.3 2.2 seen below the bottomset beds with evidence of A9 15-25 26.0 7.4 2.4 uneven truncated beds and relict channels. A wood A9 15-40 41.0 8.1 7.3 A91 5-75 76.0 8.7 2.1 fragment from the core catcher at Station 1 was A915100 101.0 9.0 2. I radiocarbon dated at A 10,000 years, suggesting A915-I30 131.0 8.0 1.7 that the underlying surface is most likely .4915-140 141.0 8.7 2.2 Pleistocene in age. Sediment accumulation rates obtained from this study indicate that the sub- ciw h A916-0 1.o 7.9 7.9 2.4 aqueous delta is an active feature. High accumulation rates are observed for the thickest (foreset) Core8 part of the clinoform, ranging from about A918-0 1.0 7.0 7.0 I.9 I- > 5 cm/year, and rates decrease to < 0.3 cm/year .4918-g 9.0 7.0 2.4 in the bottomset region. Although accumulation .4918-15 16.0 6.9 2.3 rates were not determined for the nearshore topset A918-25 36.0 6.9 2.3 region, the abundance of coarse (sandy) sediment C‘ore 9 indicates bypassing of fine-grained sediment to the A919-0 1.0 8.6 7.6 3.3 middle and outer-shelf areas. A919-2 3.0 7.4 2.1 A919-IO 11.0 9.0 0.7 5.2. Slw(f’sedinzent disprrsul A919-15 16.0 8.4 I.2 A919-45 46.0 5.6 2.6 A919-50 51.0 6.7 2.7 Textural and acoustical characteristics of the seabed provide clues to patterns of shelf sediment dispersal. In the nearshore region, poor acoustic are observed and field descriptions of grab samples penetration (O-25 m) and our inability to collect from this area reveal a clean medium sand. The long gravity cores suggest a sandy seabed. Along sand waves have considerable relief (3-5 m), the eastern margin, near Chittagong, sand waves implying a high-energy environment. The orienta- S.A. Kuehl et al. 1 Marine Geology 144 ( 1997) 81-96

Fig. 11. Cartoon illustrating features of the subaqueous delta on the Bengal shelf seaward of the Ganges-Brahmaputra river system. The clinoform is characterized by relatively coarse-grained (sand) topset beds and fine-grained (mud) foreset and bottomset beds. Evidence from a variety of sources indicates that a significant fraction of sediment discharged to the shelf is transported seaward and westward along the shelf, and escapes to the deep sea through the Swatch of No Ground. tion of lee and stoss sides of the sand waves accumulation of fine-grained sediments near shore. indicates east to west flow direction (Fig. 5). These Westward sediment transport is evidenced by fining features disappear toward the west, and acoustic towards the west, sand wave orientation, and penetration gradually increases as the sediment increased sedimentation rates near the Swatch of becomes finer. Along the middle shelf, textural No Ground. This seabed evidence is corroborated analyses suggest a fining trend westward from by physical oceanographic observations and mod- Station 8 (7.0 4) toward Station 6 (7.9 4) and 5 eling which indicates westward extension of the (8.5 4). near the Swatch of No Ground. Surface river plume during the southwest monsoon (Shetye grain-size distributions from grab samples demon- et al., 1996). strate a similar pattern of westward fining (Kuehl The Swatch of No Ground appears to divert et al., 1989). The highest acoustic penetration is the westward dispersal of sediments from the shelf observed near the canyon (> 100 m) indicating into the canyon. This is supported by clay mineral- thick deposits of fine-grained sediments. ogical studies which show a depletion of chlorite The evidence above indicates southward and on the western side of the canyon, indicating that westward transport of fine-grained sediments from transport of chlorite-rich Brahmaputra sediment the rivers’ mouths toward the Swatch of No is interrupted by the canyon (Segall and Kuehl, Ground. Energetic waves and currents at times of 1992). The Swatch of No Ground incises the maximum sediment discharge likely prevent rapid middle shelf in the western foreset region, the area of highest measured sedimentation rates Holocene sediments deposited seaward of the max- (> 5 cm/year). Evidence of mass movement is imum transgressive shoreline. common in areas surrounding the canyon. High The Khulna region is considered the northern- rates of sediment loading are one cause of sediment most extent of the Ganges-Brahmaputra delta failures such as growth faulting, slumping, and shoreline during the Holocene (Umitsu, 1993 ). mud flowage (Figs. 9 and 10). These observations Two fossil molluscs, Corhiculidue geloinu and are consistent with earlier studies showing rapid Neritiolue neritinu, found at a depth of 16 m sug- accumulation and penecontemporaneous deforma- gest that this was once a tidal environment. tion of sediment fabric in the vicinity of the canyon Radiocarbon dating of wood fragments found in ( Kuehl et al., 1989, 1991). The existence of growth the core at a depth of 16 m below sea level reveal faults and slumps directed into the canyon. coupled an approximate age of 7000 14C years before with the irregular. thick sequences of sediment present ( Umitsu, 1993 ). The 7000-year shoreline found at the bottom of the canyon, is dramatic is extrapolated laterally from Khulna for the evidence that some sediment is bypassing the shelf volume estimates. In the east, the mountainous via the Swatch of No Ground. presumably to the Chittagong hill tract provides a natural boundary. Bengal Fan. A southern boundary is provided by the 80-m Evidence of mass movement is also observed in isobath; bottomset beds are thin at this water the relatively steep foreset region of the subaque- depth and immediately overly the Pleistocene sur- ous delta where two distinct, acoustically transpar- face. The western boundary is taken to be the ent layers are observed from the GeoPulse@ Hooghly River, just to the west of the Swatch of profiles (Fig. 7). The transparent layers extend No Ground, which was a major outlet for the from the outer topset to inner bottomset beds, Ganges River in recorded history. Bathymetric following the gradient of the subaqueous delta. and topographic data were used to constrain the and have an area) extent of _ 1500 km’. One upper surface. and the lower surface was taken as possible explanation for these features is that they a plane extrapolated landward from the -80-m represent large-scale mud flows similar to those isobath. All depth/elevation and position informa- observed off the Mississippi River (Wright and tion was processed using Surfer@ to calculate the Coleman, 1974). The widespread nature of these volumes for the marine ( 1.97 x IOr m3) and ter- features suggests large-magnitude external forcing restrial (0.26 x 10” m3) sediment accumulation for failure. Such forcing could be accomplished by seaward of the Holocene maximum transgression. cyclones or earthquake activity common in the Based on the estimated annual sediment load study area. of 1060 x lo6 t/year ( Milliman and Syvitski, 1992). the volume of marine material on the shelf (integrated over 7000 years and assuming a dry bulk 5.3. Volume oj’Holocene sediment w!edge density of 1.1 g cm-3) represents about 3 1% of the rivers’ input. The corresponding terrestrial portion In order to examine the significance of subaque- is about 4.2%. Error of these budget estimates ous deltaic progradation for the Ganges- likely is large as they are based on modern sediment Brahmaputra system since the maximum Holocene discharge figures that may not be representative transgression. a first-order estimate was made of of the entire late Holocene. For example. anthro- the volume of the sediment wedge forming on the pogenic activities can have significant effects on inner shelf. Area1 boundaries were delineated sediment discharge; farming can increase load primarily based on seismic data from this whereas river-control structures, such as dams, can study, stratigraphic analysis and interpretation of reduce the load through trapping. Uncertainty in Umitsu ( 1993), and geographic information. the chronology taken from Umitsu ( 1993) contrib- Extrapolation of Umitsu’s bore-hole data to the utes further to the potential error of this estimate. offshore enables estimates for the thickness of Most importantly, the boundaries were based on subaerial and subaqueous components of sedimentary facies and geographical features and S.A. Kuehl et al. / Marine Geology 144 (1997) 81-96 95 are, as such, arbitrary to some degree. Despite seaward and westward, with the thickest accumula- these uncertainties, the subaqueous delta off the tion of mud near the submarine canyon, Swatch Ganges-Brahmaputra clearly represents a signifi- of No Ground, which incises the western shelf. cant component of the late Holocene delta. Sand waves found near the eastern coast of In comparing the integrated discharge values to Chittagong are oriented with their lee side facing the volume of the subaqueous delta, an important west, also indicating westward transport. consideration is that river discharge measurements (3) Slumps, growth faults and evidence of mass are taken some distance upstream of the shoreline movement coupled with high sedimentation rates and may not reflect the amount supplied to the near the Swatch of No Ground provide dramatic coastal ocean, Floodplain sedimentation appears evidence that modern sediment is being channeled to be a significant factor, accounting for a loss of off-shelf through the submarine canyon to the lo-40% of the measured discharge for the Bengal Fan. Mississippi, Amazon and Changjiang rivers (Kesel (4) Sediment volume estimates reveal that the et al., 1992: Nittrouer et al., 1995; Xiqing, 1996). Holocene subaqueous delta accommodates about For the Ganges-Brahmaputra system, estimates of 30% of the rivers’ load, indicating that subaqueous floodplain sequestering range as high as 80% of deltaic progradation is an important sink for the the discharge, although these estimates are based rivers’ sediment. The remainder is partitioned on a limited data set (Milliman and Syvitski, between the floodplain/delta plain and deep sea, 1992). Because floodplain accommodation could but the relative importance of these sinks is not significantly reduce the rivers’ input to the shelf, known. the proportion of sediment accumulation repre- sented by the subaqueous delta could easily exceed the above estimate of 31%. A recent study of Acknowledgements floodplain sedimentation rates for a 110 km reach of the Brahmaputra left bank estimates the The authors would like to thank the Netherlands removal of about 5% of the river’s load (Allison sponsored Land Reclamation Project for their et al., 1998); however, extrapolation of this figure logistical support during the field work. Financial to the rest of the floodplain and delta plain is support was provided by the National Science problematic as considerable regional variation in Foundation grants OCE-9019472 and OCE- floodplain sedimentation rates in this dynamic and 9322254. Charles A. Nittrouer generously provided tectonically complex region seems likely. More access to his GeoPulse@ system used in the field detailed study is needed to ascertain the role of study. Steven L. Goodbred drafted Fig. 1. This the floodplain sedimentation in the overall budget. paper forms contribution No. 2069 from the Virginia Institute of Marine Science, College of William and Mary. 6. Conclusions

( 1) Seismic reflection profiles of the sediment References wedge seaward of the Ganges-Brahmaputra river system reveal clinoform stratigraphy characteristic Alexander, C.R., DeMaster, D.J., Nittrouer, C.A.. 1991. Sedi- of a subaqueous delta. Sediment accumul- ment accumulation in a modern epicontinental-shelf setting: ation rates are highest in the foreset region The Yellow Sea. Mar. Geol. 98, 51-72. ( l-> 5 cm/year) and decrease in the bottomset Allison, M.A.. 1997. Historical changes in the Ganges-Brahma- region (< 0.3 cm/year), an observation consistent putra delta front. J. Coastal Res., in press. with an actively prograding clinoform. Allison, M.A., Kuehl. S.A., Martin, T., 1998. The importance of floodplain sedimentation for river sediment budgets and (2) Seabed evidence suggests that the dominant terrigenous input to the oceans: Insights from the Brahmapu- transport of fine-grained sediment on the Bengal tra-Jamuna River. Geology, submitted. shelf is to the south and west. Sediments fine Barua. D.K., Kuehl. S.A.. Miller, R.L.. Moore, W.S., 1994. Suspended sediment distribution and residual transport in Nittrouer, C.A., Kuehl, S.A.. Sternberg, R.W.. Figueiredo, the coastal ocean off of the GangessBrahmaputra river .A.G.. Faria. L.E.C., 1995. An introduction to the geological mouth. Mar. Geol. 120, 41-61. significance of sediment transport and accumulation on the Barua. D.K., Martin. T.C.. Hasan, A., Mahmood. S.. Kuehl. Amazon continental shelf. Mar. Geol. 125, 1777192. S.A.. 1998. Suspended sediment transport through the Gan- Polach. H.A.. Stipp. J.J.. 1967. Improved synthesis technique gessBrahmaputra fluvial system: Implications for floodplain for methane and benzene radiocarbon dating. Int. J. Appl. sedimentation and sediment delivery to the ocean. Sediment. Radiat. Isot. 18, 3599364. Geol., submitted. Prior, D.B., Yang. Z.S.. Bornhold. B.D.. Geller, G.H., Lin, Coleman, J.M., 1969. Brahmaputra River: Channel processes Z.H.. Wiseman Jr.. W.J.. Wright. L.D., Lin, T.C.. 1986. The and sedimentation. Sediment. Geol. 3, 139 -239. subaqueous delta of the modern Huanghe t Yellow River). Cutshall. N.H., Larsen. I.L.. Olsen, C.R.. 1983. Direct analysis Geo-Mar. Lett. 6. 67-75. of “‘Pb in sediment samples: Self-absorption correction. Segall. M.P.. Kuehl, S.A.. 1992. Sedimentary processes on the Nucl. Instrum. Methods 206. 309- 3 I?. Bengal shelf as revealed by clay-size mineralogy. Cont. Shelf Geyer. W.R.. Beardsley. R.C., Lentr. S.J.. Candela. J.. Res. 12, 517 541. Limeburner, R.. Johns. W.E., Castro. B.M.. Soares. I.D.. Segall. M.P., Kuehl. S.A.. 1994. Sedimentary structures on the 1996. Phys. Oceanogr. Amazon Shelf 16. 575 616. Bengal shelf: a multi-scale approach to sedimentary fabric Kesel, R.H.. Yodis, E.G., McGraw, D.J., 1992. An approxima- interpretation. Sediment. Geol. 93. 165-180. Shetye. S.R.. Gouveia. A.D.. Shankar, D., Shenoi, S.S.C.. tion of the sediment budget of the lower Mississippi river Vinayachandran. P.N., Sundar. D.. Michael, G.S., Nam- prior to major human modification. Earth Surf. Process. poothiri. G.. 1996. Hydrography and circulation in the west- Landf. 17, 711-722. ern Bay of Bengal during the northeast monsoon. Kuehl, S.A.. DeMaster, D.J., Nittrouer. C.A.. 1986. Nature 01 J. Geophys. Res. 101. 1401 ll14026. sediment accumulation on the Amazon continental shelf. Suryanarayana. A.. Murty. C.S., Rao. D.P.. 1992. Characteris- Cont. Shelf Res. 6, 209 225. ttcs of coastal waters of the western bay of Bengal during Kuehl. S.A., Hariu. T.M.. Moore. W.S., 1989. Shelfsedimenta- different monsoon seasons. Austr. J. Freshwater Res. 43. tion off the GangessBrahmaputra river system: Evidence for 1517 1533. sediment bypassing to the Bengal fan. Geology 17. llmitsu. M.. 1985. Natural levees and landform evolution in 1132~1135. the Bengal lowland. Geogr. Rev. Jpn. 58. 1499164. Kuehl, S.A.. Hariu. T.M.. Sanford. M.W., Nittrouer. C.A.. Ilmitsu. M.. 1987. Late Quaternary sedimentary environment DeMaster. D.J.. 1991. Millimeter-scale sedimentary struc- and landform evolution in the Bengal lowland. Geogr. Rev. ture of fine-grained sediments: Examples from continental Jpn. 60, 164 178. margin environments. In: Bennett, R.H.. Bryant, W.R.. Llmitsu, M., 1993. Late Quaternary sedimentary environments Hulbert. M.H. (Eds.). Microstructure of Fine-Grained Sedi- and landforms in the Ganges delta. Sediment. Geol. 83. ments. Springer, New York, pp. 33 45. 177 186. Kuehl. S.A.. Pacioni. T.D., Rine. J.M.. 1995. Seabed dynamics Wright. L.D.. Coleman. J.M., 1973. Variations in morphology of the inner Amazon shelf: temporal and spatial variability of major deltas as functions of ocean wave and river dis- of surficial strata. Mar. Geol. 135. 283- 302. charge regimes. Bull. AAPG 57. 370-398. Milliman, J.D., Syvitski, J.P.M.. 1992. Geomorphic, tectonic Wright. L.D.. Coleman, J.M.. 1974. Mississippi River mouth control of sediment discharge to the ocean: The importance processes: Effluent dynamics and morphologic development. of small mountainous rivers. J. Geol. 100. 5% 544. J. Geol. 82. 751-778. Mooley. D.A., Mohile. C.M., 1983. A study of cyclonic storms Wright. L.D., Nittrouer. C.A., 1995. Dispersal of river scdi- on the different sections of the coast around the Bay of ments in coastal seas: Six contrasting cases, Estuaries 18. Bengal. Mausam ( India) 43. I3991 52. 494508. Murty. V.S.N.. Sarma. Y.V.B.. Rao. D.P.. Murty. C‘.S.. IY92. Wyrtki. K.. lY73. Physical oceanography of the Indian Ocean. Water characteristics. mixing and circulation in the bay of In: Zeitzschel, B. ( Ed.). Ecology Studies. The Biology of the Bengal during southwest monsoon, J. Mar. Res. 50,207- 228. Indian Ocean. Springer, New York, pp. 18836. Nittrouer. C.A.. Kuehl, S.A.. DeMaster. D.J.. Kowsmann. Xiqing. C.. 1996. An integrated study of sediment discharge R.O.. 1986. The deltaic nature of Amazon shelf sedimenta- from the Changjiang River, China, and the delta develop- tion. Geol. Sot. Am. Bull. 97. 444458. ment since the mid-Holocene. J. Coastal Res. I>. 26- 37.